sagenb update to 1.0

[dimpase]

This is sagenb update, incorporating changes accumulating since May 2016.

Tarball: ​http://users.ox.ac.uk/~coml0531/sage/sagenb-1.0.1.tar.bz2

[dimpase]

please review!

[jhpalmieri]

See also #22787: do not build SageNB if SAGE_PYTHON3=yes. If this version fixes the Python 3 issues, then we should undo the change in #22787.

[dimpase]

Replying to jhpalmieri:

See also #22787: do not build SageNB if SAGE_PYTHON3=yes. If this version fixes the Python 3 issues, then we should undo the change in #22787.

I propose we postpone dealing with this particular issue until Sage runs on python3.

[dimpase]

I'd like to mention that after a first install of the new sagenb I sometimes have

sage -t --long --warn-long 54.6 local/lib/python2.7/site-packages/sagenb/notebook/worksheet.py
**********************************************************************
File "local/lib/python2.7/site-packages/sagenb/notebook/worksheet.py", line 477, in sagenb.notebook.worksheet.Worksheet.__lt__
Failed example:
    W1 <= W2
Expected:
    True
Got:
    False
**********************************************************************
File "local/lib/python2.7/site-packages/sagenb/notebook/worksheet.py", line 479, in sagenb.notebook.worksheet.Worksheet.__lt__
Failed example:
    W2 <= W1
Expected:
    False
Got:
    True

---something that I cannot reproduce after actually using sagenb. In fact even the re-run of the same test does not reproduce it. Not 100% sure how harmless these failures are. (I think they are harmless).

[jdemeyer]

Test failures are never harmless.

[dimpase]

If I reinstall the package then the test does not fail. Good luck debugging this, but I am not doing it.

As a workaround one can add running this test once in spkg-install and ignore its result.

[dimpase]

Has anyone seen this before? I stopped paying much attention to sagenb in 2012, so I cannot recall anything.

[chapoton]

hmm, comparison of worksheet has been changed recently (to avoid using cmp)

This lives in sagenb/notebook/worksheet.py

I think at least that maybe there lacks __ne__ and maybe also @total_ordering ?

[chapoton]

I just made a pull request on github

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​664dcc9	Merge branch 'u/dimpase/sagenb10' of trac.sagemath.org:sage into sagenb10
​70fa3f8	updates to properly compare worksheets

[dimpase]

OK, thanks to Frederic, the ordering of the worksheets is now fixed, and the doctests all pass too. (and the tarball is updated with these fixes, along with the checksum).

[jhpalmieri]

This works for me. I haven't tested that thoroughly, but all tests pass, tests do not create the "SAGE_ROOT/home" directory any more, and the notebook seems to work. "Live" help works, too. Anyone object to a positive review?

[jdemeyer]

Let me just check 2 things:

1. A rebuild from scratch (i.e. after make distclean)

2. That the tarball is correctly produced.

[jdemeyer]

Replying to jdemeyer:

Let me just check 2 things:

1. A rebuild from scratch (i.e. after make distclean)

Doing this right now...

2. That the tarball is correctly produced.

This is not looking good. Unless I'm missing something, the fixes mentioned in 11 are not in version 1.0 of SageNB. So the version number is misleading. Current SageNB master should be version 1.0.1 and then that tarball should be in Sage.

[dimpase]

right, I shall bump up the version to 1.1.

[jdemeyer]

If you want to use semantic versioning (not saying that you should), this is a bugfix, so it should be version 1.0.1

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​9f4c79c	Merge branch 'u/dimpase/sagenb10' of trac.sagemath.org:sage into sagenb101
​3ce1714	bumping up the version, change to github tar.gz file

[dimpase]

here is the bumped up version, repackaged so that the tarball is made by github.

[jdemeyer]

A git repository is not a source distribution. I suggest to use the dist.sh script to create the tarball.

That being said, it might be possible to install from a git repository, but then

1. it should be clearly documented in SPKG.txt.

2. the filename should be fixed (1.0.1.tar.gz is not acceptable because it doesn't contain sagenb).

[git]

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

​7b85dc2

bumping up the version

[dimpase]

OK, I have followed your suggestion and used dist.sh now.

PS. Regarding the somewhat unfortunate naming scheme for github "releases", I can only say that the tar.gz file gets renamed to the right one if you download it using a browser.

[jhpalmieri]

This also passes tests and works with my (cursory) usage after I built from scratch.

[jdemeyer]

Replying to dimpase:

I can only say that the tar.gz file gets renamed to the right one if you download it using a browser.

My browser (GNU wget) didn't do that.

[jdemeyer]

I'm testing this again...